Odo: Depth-Guided Diffusion for Identity-Preserving Body Reshaping

TL;DR

This paper introduces Odo, a diffusion-based method for realistic human body shape editing guided by semantic attributes, supported by a large-scale dataset for training and evaluation.

Contribution

The work presents the first large-scale dataset for controlled human shape editing and proposes Odo, a novel diffusion-based approach combining shape guidance and appearance preservation.

Findings

Achieves per-vertex errors as low as 7.5mm, outperforming baselines.

Produces realistic, identity-preserving body reshaping results.

Outperforms prior methods in accuracy and visual quality.

Abstract

Human shape editing enables controllable transformation of a person's body shape, such as thin, muscular, or overweight, while preserving pose, identity, clothing, and background. Unlike human pose editing, which has advanced rapidly, shape editing remains relatively under-explored. Current approaches typically rely on 3D morphable models or image warping, often introducing unrealistic body proportions, texture distortions, and background inconsistencies due to alignment errors and deformations. A key limitation is the lack of large-scale, publicly available datasets for training and evaluating body shape manipulation methods. In this work, we introduce the first large-scale dataset of 18,573 images across 1523 subjects, specifically designed for controlled human shape editing. It features diverse variations in body shape, including fat, muscular and thin, captured under consistent identity, clothing, and background conditions. Using this dataset, we propose Odo, an end-to-end diffusion-based method that enables realistic and intuitive body reshaping guided by simple semantic attributes. Our approach combines a frozen UNet that preserves fine-grained appearance and background details from the input image with a ControlNet that guides shape transformation using target SMPL depth maps. Extensive experiments demonstrate that our method outperforms prior approaches, achieving per-vertex reconstruction errors as low as 7.5mm, significantly lower than the 13.6mm observed in baseline methods, while producing realistic results that accurately match the desired target shapes.